REUNIÓN ESTUDIOS REGIONALES

Transcription

1 REUNIÓN ESTUDIOS REGIONALES El Estado Autonómico a debate: eficiencia, eficacia y solidaridad Alcalá de Henares, de noviembre de TITULO DE LA COMUNICACIÓN: Competitiveness and territorial cohesion in the Spanish provinces: An urban approach AUTORES: Asunción López López y Federico Steinberg Wechsler FILIACIÓN: Universidad Autónoma de Madrid CONTACTO (DIRECCIÓN E ): federico.steinberg@uam.es RESUMEN This paper presents a simple model that explores the relationship between competitiveness and territorial cohesion, and applies it to the case of Spanish provinces for the period We find that redistributive policies in Spain have been effective in promoting territorial cohesion in a large number of provinces. However, cohesion policies have not succeeded in increasing productivity, particularly in the relatively impoverished southern regions. Moreover, we show that the traditional division between dynamic and backward provinces is being overcome and does not reflect any more the economic-territorial map of Spain. After reviewing the evolution and interaction of a variety of socio-economic variables - employment, urbanization, subsidies, gross value added, disposable income, ageing, population density, and productivity - the paper concludes with some policy recommendations. PALABRAS CLAVE (DE TRES A CINCO): competitiveness, cohesion, productivity, redistribution. CÓDIGOS JEL: O18, R10, R11, R19

2 I. Introduction Both competitiveness and territorial/social cohesion are public policy goals. An optimal policy mix should not only attract investment and promote the competitiveness of firms within a territory, but also foster some degree of social and territorial cohesion. Therefore, the challenge governments face is to use state intervention in the economy in a way that maximizes the possibilities of achieving these goals simultaneously. However, if government intervention is insufficient, excessive or wrongly targeted, it could well reduce the competitiveness and growth prospects of regions, and even handicap cohesion on the long run. Therefore, it is important to understand which public policies that aim at promoting social and territorial cohesion have positive effects on competitiveness and which ones fail to do so. Examples of such potentially beneficial policies include income redistribution through taxes, subsidies, and pensions; public spending in infrastructure, urbanization programs, education and training programs, or labour market reforms. A better understanding of the specific positive and negative externalities of these public policies is necessary to envisage what kind of governance actions are required to promote both competitiveness and territorial/social cohesion. Moreover, the increase in economic interdependence, usually referred as globalization, makes it even more important today to explore the truly relationship between these variables, which is still poorly understood. In this paper, we propose a simple model that explores the relationship between competitiveness and cohesion. We then apply it to the case of Spanish provinces for the period , which corresponds to the creation of the decentralized Estado de las Autonomías. Our quantitative analysis captures the effects of a variety of public policies aimed at redistributing income on the evolution of production, productivity growth, and employment. It also explores the evolution of urban concentration and its effect on the distribution of population and disposable income across regions. We find that redistributive policies in Spain have been effective in promoting territorial cohesion in a large number of provinces. However, cohesion policies have not succeeded in increasing productivity, particularly in the relatively impoverished southern regions. The catch up process of the backward regions and the general convergence process (in terms of both per capita disposable income and urbanization) that has characterized Spanish provinces since the 1980s has been primarily the result of population movements and urbanization. Subsidies to firms and direct transfers have only had a marginal positive impact, and only in some provinces. In fact, we find that some regions have been unable to increase productivity and employment at sustainable levels even though they have received substantial transfers. Within these general observations, the paper presents new relevant conclusions by grouping provinces into five different categories and by analyzing a variety of competitiveness and cohesion indicators over time. The structure of the paper is the following. Section II proposes a general theoretical framework and reviews the relevant literature, focusing specifically on the empirical works conducted for Spanish regions. Section III defines the key variables and presents the model. Section IV discuses the main results and section V highlights the conclusions.

3 II. General framework and literature overview Conceptual framework: what role for the state? According to neoclassical economics, market forces would automatically create a virtuous circle in which competitiveness and social cohesion reinforce each other and in which different territories tend to converge. The logic of the argument is represented in figure 1. Figure 1 Competitiveness Productivity increase GDP increase Investment Welfare increase Social cohesion A country with high levels of social and territorial cohesion (which imply well functioning institutions, political and economic stability, a balanced distribution of income, and a favourable business environment) would attract capital because it will offer investors relatively high expected returns. Through the process of capital accumulation, productivity will raise, which in turn, will increase the competitiveness of the region. This economic dynamic will increase GDP in each region and, provided that this larger pie is distributed to all economic actors, welfare will rise, strengthening social cohesion. 1 Moreover, the process should be reinforced by the tendency of income per capita to converge, at least in relative terms. 2 Historical experience, however, has shown that this process is not automatic. Markets cannot bring about this virtuous circle because there are several market failures that require government intervention. In fact several highly competitive regions in the United States (and also in Europe) rank among the least cohesive in terms of economic equality, wage differentials and crime rates. Therefore, the public sector has to perform several functions, namely: 1 The origins of this idea can be traced back to Smith (1776). Contemporary liberal scholars, who support minimum intervention of the State in the economy, are the main supporters of this idea. However, they tend to give relatively more weight to competitiveness than to cohesion. 2 This result, which remains controversial, is one of the basic conclusions of neoclassical growth models. For a review of both neoclassical and endogenous growth models, see Barro and Sala-i-Martin (2004).

4 o Provision of public goods: the government is the only actor capable of increasing physical capital (infrastructures) and human capital (education) which are necessary to attract investment and to make sure that capital accumulation effectively translates into higher total factor productivity. o Income redistribution: there is not a direct link between GDP increase and real welfare increased (measured by an increase in real income and a relatively equal income distribution). In order for welfare to increase, the public sector has to redistribute income through the tax system, transfers and pensions. o Social infrastructure: it is important to point out that the government has to ensure that the institutional framework is favourable. It has to protect property rights and contracts, provide security and stability, and facilitate dialogue among stakeholders (business and labour) in order to consolidate social cohesion. 3 The analytical scheme presented in figure 1 can therefore be tentatively extended in the following way to show where public sector intervention is necessary taking into account the regional dimension of redistribution (see figure 2). Figure 2 Competitiveness Productivity increase GDP increase in the most dynamic regions Income redistribution at the regional level Investment Provision of infrastructure and education Market forces Public sector intervention Social/territorial cohesion Balanced welfare increase in different regions Institutions for conflict resolution The logic of this general framework is usually supported by scholars aligned with social democracy and/or with institutional approaches. However, it is difficult to empirically test its validity because in the real world it is virtually impossible to isolate the influence of each variable on others, even with sophisticated econometric techniques (Rodrik 2005). Therefore, most studies have taken only small parts of figure 2 and studied the interaction of some variables in isolation. 3 For a detailed technical discussion of the role of the state in performing these functions see Stiglitz (1999) and Rosen (2002). On the role of institutions on economic growth and development, see North (1990).

5 Figure 2, however, allows us to identify two different sets of questions. The first has to do with the effects of public spending on territorial cohesion and income distribution: Are national income disparities reduced over time when the state redistributes income? Is there convergence in regional and/or personal disposable income per capita? The second set of questions, which are more complex, refer to the degree of effectiveness of redistributive policies in increasing productivity: Is income redistribution a useful tool to increase the productivity and competitiveness of the more back warded regions? Is the process of catch up predicted by neoclassical growth models strengthen or undermined by a strong state intervention focused on redistributing income? And finally, are competitiveness and cohesion compatible or incompatible policy goals? In the following pages we try to respond these questions. Literature Overview Before we develop our model and main results, we briefly review the relevant literature. We focus more on the empirical studies conducted for Spanish regions than on general theoretical issues. A discussion of theoretical concepts in the context of the European integration process can be found in Rodriguez Pose (1998). The pioneer theoretical work on the incidence of public spending on income distribution corresponds to Musgrave (1959). His work on public economics triggered both theoretical and empirical work on the role of the state in the economy, including issues of income redistribution. 4 The most important empirical contributions, which had a significant influence in subsequent research, are Gillespie (1965) and Musgrave et al. (1974). Since the 1970s the relevance of the theoretical research has diminished. Contrarily, the number of empirical works has grown exponentially and income redistribution has been address at the personal, spatial, sectoral, and functional levels. For the Spanish case, Molina (1983) undertook the first empirical relevant study on the incidence of public spending on income distribution. Most of the subsequent relevant conclusions can be found in the book edited by Mella (1998). One of its key findings is that the public sector has contributed to increase per capita income in some backward regions through transfers. Within the most recent research Parellada (2003) finds that, within the spectacular economic growth of the Spanish economy since the 1960s, there has been an important degree of convergence in per capita income, but also a significant degree of divergence in the territorial distribution of production. He also finds that the reductions in spatial income inequality, especially for the periods , are explained by the fact that factors of production have moved in net terms from the regions with relatively less income to those with relatively more income, which also explains that production has concentrated in specific areas. 5 Deepening in the issue of income distribution Goerlich and Mas (2004) find that Spain became more equal between 1973 and 1991 (the national Gini coefficient went form to ) but that inequality increased between 1991 and 2001 (the Gini coefficient for 2001 was ). They also found that inequality 4 Adler (1951) Tucker (1954), and Conrad (1954) had conducted some previous empirical analysis. 5 Specifically, there is a relative decline in Asturias, Cantabria and País Vasco, and, to a lesser extent, Castilla la Mancha, Castilla León, and Extremadura. On the other hand, Madrid, Baleares, the Canary Islands and the Meditarranean regions improve their relative positions.

6 increased mainly because the richest 5% of the population significantly increased its share in total income during the nineties. They also point out that the richest regions tend to be the more egalitarian. In a different study, Villaverde Castro (2001) finds that for the period convergence on the regional distribution of GVA per capita has been limited. He explains this fact by the increasing inequality on occupational rates. He also argues that the differences in productivity across regions are explained by the different structural productive systems that the different regions have, and recommends labour market reforms to foster a process of convergence in regional per capita income. III. The model We propose a model that analyzes the evolution of set of socio-economic variables over time and allows us to capture the dynamics and interaction of territorial cohesion and competitiveness. We start by presenting some basic definitions for the key variables that we will examine: Competitiveness: the capacity of a province to generate wealth. Wealth is measured in terms of production per capita, productivity, and population dynamism. Territorial cohesion: increase in homogeneity in the inter-provincial distribution of net household disposable income per capita (NHDI pc). Starting form a relative unequal income distribution, a process of territorial cohesion requires interprovincial income transfers. We measure these income transfers evaluating the difference between GVA and NHDI. The main hypotheses we will test are: The territorial competitiveness map is changing in Spain. The traditional difference between advanced and backward provinces slowly disappearing. The decentralized system of the Estado de las Autonomías has not been able to globally contribute to the reduction of income disparities. Our analysis uses data published by Fundación BBVA (2003). The variables considered are Total employment Gross value added a factor cost Unemployment rate Total population Non-agrarian employment Net household disposable income Index of implicit prices in the Gross Active population Value Added. All Sectors (1986 = Employed population 100) Population between 16 and 64 Urbanization rate Population over 65 Subsidies Population over 15 The analysis has been conducted for the period In 1981 the Estado de las Autonomías (territorial division in 17 Autonomías or regions and 50 provinces) and the Welfare State were established in Spain is the last year for which all the variables are available. Given that there are various ways to refer to the variables we use, it is necessary to define some of them specifically.

7 Net household disposable income: the value of all income earned by households and non-profit private institutions from salaried labor and returns form capital (including rents for those owners that use their own houses as work places), usually referred as Direct Rents. We add social spending and other transfers to households and subtract direct taxes and social security contributions paid by households and business. In sum, it is the net income available by households to consume and save (BBVA, 2003, p.16-17). Urbanization index: percentage rate of the population that lives in urban areas in each province. Urban areas in each province are taken from the definition of the Ministerio de Fomento (Ministerio de Fomento, 2003) Subsidies: amounts received either from the domestic public sector or from foreign institutions. Their primary goal is to foster the development of productive economic activities. They include the transfers that farmers receive form the CAP. Methodology The evolution of competitiveness is calculated taking into account that the nominal value added per capita in each province (Vp) can be written as a function of productivity (Prod) and the ratio of employed population over total population (Oc) (De la Fuente, Vives, 2003). GVA GVA O Vp = = = Pr od Oc (I) P O P Where GVA represents Gross Value Added, P represents total population and O represents employed population. It is possible to further decompose (I) taking into account that Prod can be expressed as: GVA Vc Pr od = = Pr ( 1+ Su) = Pr( 1+ Su) Vco (II) O O Where Pr represents the evolution of production prices, Su is the weight of subsidies in the nominal GVA and Vc is the GVA in constant terms and net of subsidies per employed person. Finally, Oc can be expressed as: O P16 64 AP O Oc = = = PPWA APWA REAP P P P16 64 AP (III) Where P16-64 is the population between 16 and 64, AP is active population, PPWA is the percentage of the population in working age, APWA is the rate of active population over the total population in working age and REAP is the rate of employment per active population. In sum, we can express nominal gross value added per cápita in one province taking logs and differences with the national average for each variable: Vp = prod + oc = ( pr + 1+ su + vco) + ( ppwa + apwa + reap) (IV)

8 The lower case letters represent the different components of Vp. We measure the evolution of territorial cohesion by calculating how income levels, (NHDI per capita) in the different provinces evolve with respect to the national average. For each variable we calculate the standard deviation and the national average. We recognize that income per capita in each province is also influenced by the transfers form the public sector (pensions, subsidies and other transfers). Thus, our calculations take into account EU transfers form CAP, unemployment rates and percentage of the population over 65 (retired). Given the heterogeneity of the different provinces, we have grouped them according to three criteria: 6 Given their relative size, big cities have been put together into a single group. This will capture their relatively high weight in the national population, GVA and NHDI. We distinguish between dynamic and backward provinces, according to the levels of NHDI per capita in The dynamic provinces are those that are above that level and the backward provinces are those that are below that level. 7 Finally, we make a distinction between the good performers and the bad performers provinces according to their participation in the national GVA. The good performers are those that gain participation in the national GVA and the bad performers are those that lose participation over time. The application of the mentioned criteria is shown in the map 1. Map 1: Provincial Grouping 6 In a different paper we will use the techniques of multivariate analysis to derive a different taxonomy of the provinces. 7 Vizcaya and Huesca have been the only two exceptions. In the case of Vizcaya the figures for 1981 hide the fact that this province has tended to be above the Spanish average in terms of NHDIpc. In the case of Huesca, even though the 1981 NHDIpc is above the national average, most of its indicators (population, aging, urbanization rate, etc) show levels similar to those of the backward provinces.

9

10 IV. Results Comments to the competitiveness tables 1.- Al table 1 shows Madrid and the good performers (both within the dynamic and backward groups) lead the growth of Spanish GVA. Barcelona loses strength as an engine of growth and the bad performers (both within the dynamic and backward groups) contribute negatively to national growth. Inter provincial differences in term of GVA per capita do not change significantly between 1981 and 1998, as shown by the fact that standard deviations are virtually the same for both years. Nevertheless, GVA tends to concentrate in the big cities and the dynamic provinces to the detriment of the backward provinces. Table 1. Distribution and Evolution of the real GVA (1986 pts) Distribution real GVA per capita GVA Annual variation (%) Big cities 29,18 29,25 730, ,36 2,91 Dynamic provinces 27,07 27,68 672,18 957,91 3,08 Good performers 19,42 20,08 683,56 956,67 3,19 Bad performers 7,66 7,61 644,96 961,21 2,83 Backward provinces 43,75 43,07 537,14 745,68 2,75 Good performers 20,09 21,30 517,70 743,94 3,42 Bad performers 23,66 21,77 554,82 747,39 2,19 Spain 100, ,00 618,53 871,52 2,89 Standard deviation 4,66 4,63 21,91 21,20 27,57 The following maps show the territorial distribution of the provinces described as good performers, bad performers, dynamic and backward. Map 3 divides provinces in a traditional way, distinguishing between dynamic and back warded provinces. Map 2, on the other hand, shows good and bad performers. A comparison of both maps shows that the regions that are improving are expanding toward the southeast and the periphery of Madrid, while those that are lagging behind tend to concentrate in the northwest, western Andalusia and some areas of the north east.

11 Map 2: New Economic Map of Spain Map 3: Traditional Economic Map of Spain 2.- As we show in table 2, in aggregate terms, the increase in population is driven by the accumulation of population in the dynamic provinces. A more detailed analysis shows that the provinces that make it possible that the Spanish population increases are the good performers (dynamic and backward). Population actually decreases in most of the bad performers, both within the dynamic and backward groups. In the big cities, Madrid shows an intense increase in population while Barcelona does not experience any significant change. The data on population density makes clear that there are important divergences across Spanish regions. Big cities and advanced provinces have increased in population density over time, while backward provinces have seen their population density decrease. A combined analysis of GVA per capita and population density in fact GVA per unit of land is just the ratio of these two variables shows that in the big cities the concentration of GVA is explained by the first variable and in the dynamic provinces

12 and the good performers within the backward provinces group it is explained by both. Therefore, the ability of big cities to attract population has played an important role in their economic dynamism. Table 2. Population: Density and variation Big cities Dynamic provinces Good performers Bad performers Population Density , ,51 Population Annual variation (%) 0,28 136, ,42 0,41 128, ,38 0,60 163, ,84 0,04 Backward provinces 64,02 63,97 0,34 Good performers 56,95 59,20 0,59 Bad performers 72,19 69,47 0,11 Spain Standard deviation 100, ,00?? 944, ,24 0,34 173, The evolution of productivity (GVA/employed person) shows a catching up effect: the most significant increases in productivity correspond to the backward provinces, followed by the most dynamic and the big cities. However, the productivity increases in the most backward provinces hide the severe unemployment problem of these areas, which tends to worsen over the period considered. Therefore, the improvements in productivity are explained by the increase in unemployment (ie. those who are employed are more productive). The slower growth in the productivity of the big cities (vis-à-vis other regions) is explained by the significant increase in employment, particularly in Madrid. In addition, the increases in population in the good performers provinces explain why productivity increases in these regions is slower than in the backward regions. Therefore the increase in productivity of the bad performers provinces is not due to stronger economic growth, but to smaller increases in employment. This result demonstrates that these provinces have not been able to break the vicious circle of economic underdevelopment. Keeping in mind this important general result, the data also shows that there has been a convergence process in productivity, which can be measured by the

13 reduction in the Standard deviation in productivity, which in 1998 is almost half of its 1981 value. Table 3. Productivity realgva per employed person (Thousands of 1986 pts) realgva per employed person Annual variation (%) Big cities ,85 Dynamic provinces ,74 Good performers ,72 Bad performers ,81 Backward provinces ,93 Good performers ,80 Bad performers ,99 Spain ,59 Standard deviation 26,00 13,41 70, Table 4 clearly shows that non agrarian employment has increased its weight in all provinces, except Cadiz. Moreover, it is above the national average in big cities and dynamic provinces. Therefore, we can conclude that: The backward provinces are those in which the weight of the agrarian sector is above the national average. In the big cities and, to a lesser extent, in the dynamic provinces, the structural economic change has taken place in the non agrarian sector. The most profound changes in the employment structure have been intra-sector, both in industry and particularly in services. These changes have been driven by the increase in both GVA and employment in the industrial and service sectors. Within the backward provinces, the structural change in favour of the industrial and service sectors has been more intense in the bad performers than in the good performers. However, since they started from a relatively more back warded level, their relative advance has not been sufficient to allow them to reach the national average level. Therefore, we can conclude that the structural change (i.e. the shift in employment toward the more productive sectors) partly explains the different levels of dynamism across regions.

14 Table 4. Non agrarian employment Non agrarian employment / Total employment (%) Annual variation (%) Big cities 98,25 99,12 0,05 Dynamic provinces 87,40 94,11 0,45 Good performers 87,32 94,18 0,46 Bad performers 87,61 93,89 0,42 Backward provinces 72,90 85,61 1,03 Good performers 75,24 86,67 0,89 Bad performers 70,99 84,60 1,13 Spain 83,35 91,60 0,58 Standard deviation 15,43 8,05 152, The rate of urbanization 8 is well above the average in the big cities, approximately on average in the dynamic provinces, and below the average in the backward provinces (table 5). However, the reduction on the standard deviation shows that there is a convergence trend because the backward provinces have experience substantial increases on their urbanization rate that have not been matched by the other provinces. This trend is explained by the fact that the backward provinces are still experiencing an urbanization process while the big cities are starting to experiencing de-urbanization and re-urbanization processes. In Madrid, the simultaneous increase in population and reduction in the urbanization rate is explained because some citizens are moving to small municipalities to reduce congestion costs. In addition, Madrid is probably witnessing a spread effect of its population to neighbour provinces like Guadalajara and Toledo. In the case of Barcelona, a similar process might be taking place, with population relocating in Tarragona and Gerona. In the bad performers the increase of the urbanization rate and, in some cases, the reduction in total population are a consequence of the process by which the smaller municipalities are loosing population while the relatively bigger cities are absolving population. This process is also taking place in the provinces of Gipúzcoa and Vizcaya. A comparison between population growth and urbanization rate shows that the latter is growing at a significant faster rate. This means that the population is increasingly concentrating on cities, which means that the demographic dynamism is clearly an urban phenomenon. Finally, a combined analysis of the population density and urbanization data, for which the highest values are exhibited by big cities and the lower values by the backward provinces, leads to the following conclusion: not only is population more 8 The rate of urbanization has been calculated using the urban areas defined by the Ministerio de Fomento.

15 concentrated in cities, but cities also tend to be bigger in dynamic regions (and obviously in big cities) than in backward provinces. Table 5. Urbanization index Urbanization index (%) Annual variation (%) Big cities 92,66 94,25 1,72 Dynamic provinces 69,32 71,60 3,29 Good performers 70,16 72,18 2,88 Bad performers 67,28 70,05 4,11 Backward provinces 59,69 64,63 8,29 Good performers 58,98 64,94 10,10 Bad performers 60,33 64,33 6,64 Spain 70,23 73,64 4,85 Standard deviation 31,19 28,24 197, Our analysis of subsidies (table 6) highlights that:: Madrid, which was heavily subsidized in 1981, still receives a significant amount of transfers 17 years later. However, Barcelona is a province that has gone from being a high recipient of subsidies in per capita terms, to receive significantly less than the national average. The dynamic provinces were receiving more subsidies than the national average in per capita terms in 1981, but received less than the national average in This change is explained by the fact that the per capita subsidies received by the good performers within the dynamic provinces group have fallen relative to the national average. This demonstrates that the growth capacity of these provinces does not depend on the transfers received as subsidies. On the other hand, it is important to point out that the bad performers within the dynamic provinces group experienced a significant increased in subsidies received per capita. However, these subsidies did not allow them to grow faster. This reveals that subsidies are not an optimal tool to promote efficient and productive economic activity. The backward provinces were receiving in 1981 less subsidies in per capita terms than the national average. The situation has dramatically change during the period considered. In 1998 they were receiving substantially more subsidies than the national average. Within this group, the per capita subsidies are higher for the good performers than for the bad performers, which suggests that for these provinces subsidies have played a positive role in increasing their growth potential. The provinces that receive higher subsidies are located in Castilla y León (Soria, Palencia, Burgos, Segovia, and Zamora), Castilla-La Mancha (Cuenca, Guadalajara, and Albacete), Aragón (Huesca and Teruel), Andalucía (Jaén), and Catalonia (Lleida). A plausible explanation for the high value in the two Castillas and Andalusia is that these regions receive a relatively high proportion of CAP funds (for cereal and oleaginous production), which are included in total subsidies.

16 On the other hand, there are some backward provinces that receive a relatively low proportion of subsidies per capita, like the ones in Galicia, Andalusia (except Jaén), and Extremadura. Finally, the standard deviation of per capita subsidies increases significantly. This reflects a more unequal distribution in favour of the good performers within the backward provinces group. Therefore, we conclude that there is no correlation between subsidies to firms and per capita income. This result is consistent with the idea that the goal of these kind of transfers is not redistributive. Table 6. Distribution and evolution of subsidies Distribution of subsidies Subsidies per capita (1986 pts) Big cities 27,76 23, Dynamic provinces 26,47 22, Good performers 17,86 14, Bad performers 8,61 7, Backward provinces 45,77 53, Good performers 18,17 27, Bad performers 27,60 26, Spain 100,00 100, Standard deviation 2,73 2,22 40,08 59,33 Comments to the cohesión tables 7.- The evolution of the distribution of the GVA and the NHDI (table 7) shows that the decentralization process of the Estado de las Autonomías in Spain has increased solidarity in some regions and reduced it in others. Specifically, those Autonomías with foral systems (País Vasco and Navarra) have reduced their territorial solidarity, which goes against the rationale and the coherence of the Welfare State. Contrarily, the four provinces of Catalonia were transferring to the other provinces in 1998 a significantly higher proportion of their GVA than in 1981, which reduces its participation in NHDI. In the case of Madrid, whose contribution so inter-territorial cohesion was (and still is) the highest in Spain, we observe a decreasing trend. The recipients of the solidarity of the big cities and of the dynamic provinces (especially some of the good performers like Alicante, Baleares, Gerona, and Zaragoza) have been the backward provinces, and particularly the bad performers within that group.

17 Table 7. Distribution of real GVA and NHDI GVA - NHDI (%) Big cities 2,20 2,16 Dynamic provinces 0,74 0,77 Good performers 0,03 0,53 Bad performers 0,71 0,25 Backward provinces 2,94 2,94 Good performers 1,68 1,32 Bad performers 1,26 1,62 Spain 0,00 0,00 Standard deviation* 0,27 0,25 Calculated with original values, not with indexes 8.- Our analysis of the NHDI per capita (table 8) makes clear that: There still are strong differences among provinces. Some provinces in Andalusia (Sevilla, Granada and Cadiz) and in Extremadura (Badajoz) show the lowest levels in NHDI per capita. In occasions, these levels are about half of the richest provinces (Gerona, La Rioaja, Álava, and Navarra). Although they receive transfers, the backward provinces have been unable to grow at the same rate as the others, both in terms of NHDI and NHDI per capita. However, we observe some convergence amongst the backward provinces that are good performers given that they show growth rates in total per capita income higher than the national average. Once again, big cities behave differently. Madrid shows a very positive result in terms of NHDI. However, due the strong population growth, the increase in per capita income is only slightly higher than the national average. Finally, the lack of variation of the NHDI per capita s standard deviation shows that the difference in income per capita between regions has persisted. Table 8. Net Household Disposable income per capita (NHDIpc) NHDIpc (thousands of 1986 pts) NHDIpc Annual variation (%) Big cities 730, ,36 2,50 Dynamic provinces 672,18 957,91 2,50 Good performers 683,56 956,67 2,35 Bad performers 644,96 961,21 2,88 Backward 537,14 745,68 2,28

18 provinces Good performers 517,70 743,94 2,57 Bad performers 554,82 747,39 2,04 Spain 618,53 871,52 2,41 Standard deviation 15,16 15,46 38, The evolution of the unemployment rates (table 9) show that: There is a significant reduction of unemployment in the big cities, particularly in Madrid. There are no significant changes in the dynamic regions and there is an increase in unemployment in the backward provinces. Interestingly, within the dynamic provinces group, the bad performers show better results than the good performers in terms of employment. The good labour market performance in Tarragona and by the disappointing evolution of employment in Álava, Alicante and Gerona explain this paradoxical result. Unemployment remains a serious challenge in backward provinces. These provinces have unemployment levels closed to 20%, with even higher levels in almost all provinces of Andalusia (Cádiz 33%, Sevilla 29%, Huelva 28%, Granada, 28%, Jaén 27%, Córdoba 27%, and Málaga 25%). Finally the differences across provinces in terms of unemployment rates have increased, as shown by the increase in the standard deviation. Table 9. Unemployment rate Unemployment rate Annual variation (%) Big cities 16,35 11,25 D1,83 Dynamic provinces 12,88 13,20 0,14 Good performers 12,11 12,77 0,32 Bad performers 14,70 14,37 0,13 Backward provinces 14,11 19,50 2,25 Good performers 16,15 19,91 1,37 Bad performers 12,47 19,10 3,13 Spain 11,68 14,79 1,56 Standard deviation 37,66 40,84 463, Aging is one of the key characteristic of Spanish demographic dynamics. Our data show that: Backward provinces have older populations. However, this trend has slowed down. Aging has been more significant in the bad performers, both in the dynamic and backward provinces. The aging process has been intense in big cities, particularly in Barcelona. In Madrid, aging has increased, but it is still below the national average. Madrid, Málaga, Murcia, Las Palmas, Santa Cruz de Tenerife, Ceuta and Melilla are the provinces with the youngest population. A plausible explanation

19 for this fact is that these provinces have received relatively high numbers of young immigrants. The slight increase in the standard deviation shows that the older population tends to be concentrated and is not homogenously distributed across regions. Given that the percentage of population over 65 is already high in the backward provinces, ageing is now increasing in the big cities and in the dynamic provinces, especially in the bad performers within this group. Table 10. Ageing rate Ageing rate Annual variation (%) Big cities 10,58 15,72 2,85 Dynamic provinces 12,16 17,38 2,53 Good performers 12,51 16,96 2,09 Bad performers 11,32 18,54 3,75 Backward provinces 12,66 17,55 2,27 Good performers 12,19 16,24 1,96 Bad performers 13,10 18,88 2,60 Spain 13,71 21,94 3,53 Standard deviation 24,73 26,73 173, Prices have evolved in a similar way in all provinces. Therefore, this variable has not played a significant role in redistributing income. However, in could be argued that this situation is especially negative for the backward provinces because its population has to accept almost the same price increases than those of the big cities and the dynamic provinces. On the other hand, if we take into account that price levels tend to be lower in backward provinces than in dynamic provinces (especially in essential goods and services like housing) one could argue that the purchasing power of disposable income is relatively higher in the backward provinces vis-à-vis the dynamic provinces and, especially, the big cities. Comments to the synthesis indicators (table 11) Income transfers, which are the key instrument to increase territorial cohesion, are originated in the big cities and in the dynamic provinces, particularly the good performers. The backward provinces, especially the bad performers are the main beneficiaries of the territorial solidarity. Income transfers received by the bad performers within the dynamic provinces group do not solve the problems of relative backwardness, insufficient production growth and low relatively low income. However, in other cases (like the good performers ) transfers have a better relative impact, both in terms of GVA and income. Solidarity does not undermine competitiveness, income level or employment growth. In fact, as shown by the evolution of big cities and good

20 performing dynamic provinces (which are net contributors), productivity has improved and not worsened. Competitiveness gains allow provinces to increase solidarity. However, foral systems are an impediment for the solidarity of some of the rich provinces. Moreover, as shown by the case of Navarra, which has a foral system, its lower degree of solidarity does not translate into more efficiency in production. The Estado de las Autonomías has not been able to correct territorial disparities in income levels. This important conclusion is demonstrated by the fact that differences between GVA and NHDI at the provincial level are not substantially modified over the period considered. However, policies of passive assistance to the productive system (subsidies to firms) do not seem to have effects neither in increasing territorial convergence. This is shown by the high volume of subsidies that the back warded bad performing provinces receive, which do not translate in improvements in production and efficiency. The economic growth potential of provinces does not seem to depend on the level of subsidies received. For example, there are provinces that improve their performance even though their share in the national distribution of subsidies decline (dynamic provinces) whereas others (backward provinces) do not improve even with a higher rate of subsidies received.

21 Table 11. Synthesis indicators Big cities M Dynamic provinces Good performers Bad performers Backward provinces Good performers Bad performers Competitiveness GVApc Productivity Demographic dynamism NHDIpc Cohesion Unemployment rate Difference between GVA and NHDIpc 1998 Variación 1998 Variación Variación 1998 Variación M M M m m m M M m m M M m C - C M M M M M M m C + C M M M M M m m C +C m M M m M M m C -C m m M = m m M R =R M m M M m M M R -R m m M m m m M R +R M = above the spanish average; m = below the spanish average. = like spanish average. C = net contributor; R = net recipient; + =improves its position as C or R; - = worsens its position as C or R; = same position as C or R.

22 Distribution of real GVA (Spain = 100) Provinces

23 NHDI pc 1981 (Spain = 100) % Provinces

24 V. Conclusions As we have seen, it is highly complex to elaborate a comprehensive analysis of competitiveness and territorial cohesion. It is difficult to synthesize all the relevant variables in definitions that are both operative and statistically quantifiable. Nevertheless, we have obtained some interesting results regarding our original hypothesis, which have economic policy implications. On the issue of competitiveness, we have detected a relatively low rate of productivity growth in big cities. This is problematic because these provinces have tended to lead the growth of the Spanish economy. A positive development, however, is that some provinces that have usually showed relatively low levels of economic growth, have recently had significant increases in GVA, productivity, and population. This means that the traditional division between dynamic and backward provinces is being overcome and does not reflect any more the economic-territorial map of Spain. In other words, the increased openness of the Spanish economy together with the structural changes associated with EU integration, particularly the single market has increased the competitiveness of some previously backward territories. These new situation also poses new challenges for the design and implementation of policies that increase regional competitiveness. An effort to increase the endowments of human, physical and technological capital is needed to consolidate the competitiveness of these regions. Maps 2 and 3 clearly show that there has been a diffusion effect of competitiveness gains towards the southeast of the Iberian Peninsula and the islands, specially the Canary archipelago. On the issue of territorial cohesion, the fact that differences in NHDIpc between the national average and each province have not changed over the period considered shows that the Estado de las Autonomías has not been successful in reducing income territorial disparities at the national level. The process of political decentralization, which was based on culture and identity demands and on the necessity to create new instruments for regional development, presents three important characteristics regarding territorial cohesion. First, the provinces of Navarra, Madrid, and the País Vasco which have relatively high levels of income have reduced their relative participation in the inter-provincial transfers system (i.e they reduced their solidarity). On the other hand, the Catalan provinces, with similar income levels, increased it. Second, the most important net contributors to inter-territorial solidarity are Madrid (which remains the first contributor), Catalonia, and some of the dynamic provinces. The recipients are most of the backward provinces. Third, although the País Vasco has a special status with more financial autonomy than other regions, it has not been able to increase its competitiveness. These conclusions suggest that the Spanish model of regional financing should be revised. The goal of this revision, which has recently started, should be to equalize the solidarity levels of provinces with equal wealth, as well as to reduce the efficiency cost derived from a lower economic growth. This means that the revision should not be bilateral (between the central government and each region) or partial because the new system has to take into account both the flow of transfers among all the regions and the fact that a reduction on the growth potential of one region negatively affects the aggregate growth rate of the Spanish economy

25 ANNEX I. Provincial grouping Population Distribution of GVA NHDIpc Provinces Variation Variation Level 81 BIG CITIES Barcelona A D M Madrid A A M DYNAMIC PROVINCES GOOD PERFORMERS Álava (Araba) A A M Alicante (Alacant) A A M Balears (liles) A A M Castellón (Castelló) A A M Girona A A M Navarra (Nalarroa) A A M Rioja (La) A A M Valencia (Valencia) A A M Zaragoza A A M BAD PERFORMERS Guipúzcoa (Gipuzkoa) D D M Lleida = = M Tarragona A = M Vizcaya (Bizkaia) D D m BACKWARD PROVINCES GOOD PERFORMERS Albacete A A m Almería A A m Badajoz A = m Córdoba A A m Guadalajara A A m Málaga A A m Murcia A A m Palmas (Las) A A m Santa Cruz de Tenerife A A m Toledo A A m Ceuta A = m Melilla A = m Burgos D A m Cáceres D A m Cuenca D A m Granada A A m Jaén = A m Salamanca D A m Soria D A m

26 BAD PERFORMERS Asturias D D M Ávila D = m Cádiz A D m Cantabria A D = Ciudad Real = = m Coruña (A) = D m Huelva A D m Huesca D D M León D D m Lugo D D m Ourense D D m Palencia D D = Pontevedra A D m Segovia D = m Sevilla A D m Teruel D D m Valladolid A D m Zamora D = m M =Above average m= Below average A= Improve D = Worsen = = Does not change

27 ANNEX II. Differences in the distribution of GVA and NHDI GVA NHDI (%) Provinces Big cities 2,20 2,16 Barcelona 0,78 0,97 Madrid 1,42 1,19 Dynamic provinces 0,74 0,77 Good performers 0,03 0,53 Álava (Araba) 0,16 0,05 Alicante (Alacant) -0,06-0,02 Balears (liles) 0,04 0,07 Castellón (Castelló) 0,03 0,16 Girona 0,03 0,14 Navarra (Nalarroa) 0,11 0,02 Rioja (La) 0,01 0,00 Valencia (Valencia) -0,29-0,07 Zaragoza 0,01 0,18 Bad performers 0,71 0,25 Guipúzcoa (Gipuzkoa) 0,16-0,07 Lleida -0,02 0,03 Tarragona 0,21 0,21 Vizcaya (Bizkaia) 0,36 0,08 Backward provinces -2,94-2,94 Good performers -1,68-1,32 Albacete -0,07-0,08 Almería -0,15-0,11 Badajoz -0,22-0,25 Córdoba -0,22-0,30 Guadalajara 0,02 0,06 Málaga -0,21-0,10 Murcia -0,14-0,13 Palmas (Las) 0,06 0,15 Santa Cruz de Tenerife 0,04 0,15 Toledo -0,08-0,05 Ceuta -0,01-0,02 Melilla -0,01-0,02 Burgos 0,05-0,01 Cáceres -0,08-0,04 Cuenca -0,05-0,07 Granada -0,30-0,18 Jaén -0,23-0,31 Salamanca -0,08 0,01 Soria 0,00-0,02 Bad performers -1,26-1,62 Asturias 0,10-0,23 Ávila -0,08-0,07 Cádiz -0,16-0,14 Cantabria 0,04-0,08

28 Ciudad Real -0,07-0,10 Coruña (A) -0,14 0,04 Huelva 0,06-0,10 Huesca 0,03-0,02 León -0,15-0,15 Lugo -0,14-0,09 Ourense -0,22-0,15 Palencia -0,01-0,05 Pontevedra -0,18-0,12 Segovia -0,01-0,03 Sevilla -0,29-0,18 Teruel 0,02-0,01 Valladolid 0,05-0,04 Zamora -0,09-0,07 Total 0,00 0,00

29 ANNEX III. Unemployment rate (%) Unemployment rate (%) Provinces Big cities 16,35 11,25 Barcelona 17,36 13,33 Madrid 15,25 9,33 Dynamic provinces 12,88 13,20 Good performers 12,11 12,77 Álava (Araba) 8,44 10,65 Alicante (Alacant) 13,95 17,77 Balears (liles) 9,81 9,53 Castellón (Castelló) 8,25 7,95 Girona 6,76 8,45 Navarra (Nalarroa) 13,03 9,01 Rioja (La) 7,55 8,81 Valencia (Valencia) 14,16 16,41 Zaragoza 13,84 9,15 Bad performers 14,70 14,37 Guipúzcoa (Gipuzkoa) 16,65 16,21 Lleida 5,09 6,78 Tarragona 11,95 9,16 Vizcaya (Bizkaia) 17,64 18,47 Backward provinces 14,11 19,50 Good performers 16,15 19,91 Albacete 17,12 17,39 Almería 11,06 17,49 Badajoz 20,39 21,57 Córdoba 18,93 27,02 Guadalajara 17,38 10,90 Málaga 21,05 25,91 Murcia 12,73 16,03 Palmas (Las) 17,29 15,86 Santa Cruz de Tenerife 15,85 17,17 Toledo 9,21 12,78 Ceuta nd 19,46 Melilla nd 17,25 Burgos 12,64 11,04 Cáceres 14,21 17,64 Cuenca 7,35 9,94 Granada 22,95 28,67 Jaén 17,08 27,25 Salamanca 11,12 20,24 Soria 5,28 7,88 Bad performers 12,47 19,10 Asturias 11,68 14,79 Ávila 7,11 13,12 Cádiz 24,93 33,21

30 Cantabria 10,13 16,43 Ciudad Real 18,44 15,98 Coruña (A) 4,82 13,09 Huelva 19,28 27,94 Huesca 7,40 6,96 León 7,71 15,24 Lugo 4,44 10,47 Ourense 5,48 13,37 Palencia 10,44 15,10 Pontevedra 7,81 13,86 Segovia 8,98 10,95 Sevilla 23,50 29,37 Teruel 5,85 7,63 Valladolid 14,02 17,41 Zamora 9,59 13,08 Total 14,36 15,72

31 References Barro, R. y Sala-i-Martin, X. (2004): Economic Growth. Cambridge MA: MIT Press. De la Fuente Moreno, A. y Vives Torrents, X. (2003): Políticas Públicas y Equilibrio Territorial, Institut d Estudis Autonòmics y Fundación BBVA. Fundación BBVA (2000): Renta Nacional de España y su distribución provincial. Año 1995 y Avances , Fundación BBVA, Bilbao. Gillespie, W.I. (1965): Effects on public expenditures on the distribution of income, en Musgrave (ed.) (1965). Goerlich, F. y Mas, M. (2004): Distribución personal de la renta en España, Papeles de Economía española nº 100, pp Jaén, M. y Molina, A. (2001): Efectos distributivos del gasto público en España Papeles de Economía Española 88, pp Mella, J.M. (coord.) (1998): Economía y política regional en España ante la Europa del siglo XXI. Akal, Madrid Musgrave, R. A. (1959): Teoría de la Hacienda Pública. Aguilar, Madrid Musgrave, R. A (ed.): Essays in Fiscal Federalism. Brookings Institution, Washington. Musgrave R. A., Case, K. E. y Leonard, H. (1974): The distribution of of fiscal burdens and benefits, Public Finance Quarterly 2, 3. Molina, A (1983): La incidencia del gasto público. Distribución personal del gasto presupuestario en España. Tesis Doctoral, Málaga. North, D. (1990): Institutions, institutional change, and economic performance. Cambridge: Cambridge University Press. Parellada, M. (2003): Distribución territorial de la renta en García Delgado J.L. (director): Lecciones de Economía Española. Thomson Civitas, Madrid. Rodríguez-Pose, A (1998): The dynamics of regional growth in Europe: Social and political factors. Oxford: Clarendon Press and New York: Oxford University Press. Rodríguez-Pose, A. y Gill, N. (2004): Is there a global link between regional disparities and devolution? Environment and Planning A, 36, 12, Rodrik, D. (2005): Why We Learn Nothing from Regressing Economic Growth on Policies. Harvard University, Mimeo. Available at: Rosen, H. (2002): Public Finance, 6th edition. New York: McGraw Hill. Smith, A. (1776/1907): An inquiry into the nature and causes of the wealth of nations. Heidelberg: Winter. Stiglitz, J. (1999): Economics of the Public Sector, 3rd edition, Norton. Villaverde Castro, J. (2001): La distribución especial de la renta en España: Papeles de Economía Española 88, pp